Controlled, Randomized Trials

While this article feels a bit simplistic, Brooks hits on an important fact: current modeling, especially economic modeling, is able to tell us very little about the world we live in. As someone who majored in economics, this is something that became more and more apparent with every class I took. Each model requires many assumptions to be able to derive anything of value from them. As compared to the dynamic world, these models hold too many variables static (which is necessary to learn from the models in an academic sense-but limits the real-world applications). Brooks points to the many randomized trials that Capital One and Google run, and asserts that government should run them as well. Brooks admits the reasons government has not thus far. And while the lack of infrastructure is a huge reason, another big reason is the current incentive structure in government. Government officials that run real studies, ask the tough questions, and publicize the results often end up alienating large swaths of the populations, which often include their constituencies. In a political world where re-election is everything, avoiding anything that may polarize your voting base is a best practice.